Finding that the axial velocities of individual drops in polydispersed turbulent dispersions with medium hold-up (as found in liquid-liquid extraction columns) cannot be predicted by known equations from the literature, methods were derived for their direct experimental determination. Assuming power-function relations between the drop diameter and the drop velocity (for a given system, hold-up and intensity of agitation), the velocities are evaluated from simultaneous measurements of local hold-ups and drop size distributions. Sets of existing data were treated in this way to correlate the actual drop velocity against drop diameter, hold-up and agitation intensity in different types of agitated extraction columns. To check the reliability of the proposed method of evaluation, new data were measured with a pilot-plant-sized Kuhni column. Together with the above-mentioned drop size distributions and local hold-ups, response curves to a tracer injection into the dispersed phase were recorded in several positions along the column axis. A method for evaluating the individual drop velocities from the changing shape of these curves and known drop size distribution was derived. Both methods delivered practically identical results so the evaluation of earlier data was justified. The procedures may be used in solutions of the population balance models to reduce the number of unknown parameters.